FACULDADE DE ECONOMIA UNIVERSIDADE DO PORTO Essays on international trade, firms’ innovation and productivity Armando Mendes Jorge Nogueira da Silva Orientador: Ana Paula Africano Sousa e Silva Co-orientador: Óscar João Atanázio Afonso TESE de Doutoramento em Economia 2010 Nota biográfica Armando Mendes Jorge Nogueira da Silva nasceu no Porto a 16 de Maio de 1962. Realizou os seus estudos secundários na escola Manuel Laranjeira em Espinho. Licenciou-se em Economia pela Faculdade de Economia do Porto (FEP) em 1985, tendose tornado assistente estagiário da FEP em Outubro desse ano. Leccionou na FEP, a tempo parcial, até Setembro de 1996 com a categoria de assistente. Entre 1990 e 1993 também desempenhou funções docentes na Universidade Lusíada do Porto e na Universidade Portucalense. Em Maio de 1996 conclui o mestrado em Economia na FEP. A partir de Janeiro de 1986 passou também a trabalhar como técnico do Departamento de Planeamento do Banco Borges & Irmão. Em Janeiro de 1989 transferiu-se para o Banco Português do Atlântico (depois B.C.P.) onde trabalhou até Março de 1999 tendo exercido várias funções técnicas, comerciais e de direcção. A partir de Novembro de 1999 passou a desempenhar funções docentes no Instituto Politécnico do Porto (IPP), na Escola Superior de Estudos Industriais e de Gestão, tendo colaborado, enquanto docente, com outras escolas do IPP: Escola Superior de Educação, Instituto Superior de Contabilidade e Administração do Porto. Entre 2000 e 2007, desempenhou ainda funções docentes no Instituto Superior de Administração e Gestão do Porto. A partir de 2009 é membro do Conselho Geral do IPP e a partir de 2010 é membro do Conselho Geral da Fundação do IPP. Após a conclusão da parte escolar do Programa de Doutoramento, apresentou os ensaios da tese em seminários na Faculdade de Economia da Universidade do Porto, bem como nos seguintes encontros científicos internacionais: “European Trade Study Group Eleventh Annual Conference” (Roma, Setembro 2009); “Conferência – Investigação e Intervenção em Recursos Humanos” (Vila do Conde, Setembro de 2009); “35th European International Business Academy Annual Conference” (Valência, Dezembro de 2009); “ Innovation and Creativity, the driving force of Europe’s future economy. Annual conference of the Chair “Leo Tindemans” (Antuérpia, Fevereiro de 2010). ii Agradecimentos Agradeço à minha mulher o amor e a compreensão pelas horas (muitas) que roubei à sua companhia de forma a poder concluir este trabalho. Agradeço aos meus orientadores Ana Paula Silva e Óscar Afonso pelos comentários, sugestões, interesse, amizade e apoio prestados. Agradeço a Mónica Costa Dias pela disponibilidade e apoio especializado nas áreas mais problemáticas dos métodos de avaliação e em especial no apoio prestado na análise de Matching que desenvolvi no último ensaio. Agradeço também a Manuel Mota Freitas pelo apoio e confiança transmitidas em fases decisivas do meu trabalho. Não quero deixar de agradecer os comentários e sugestões que recebi ao longo dos últimos meses e que só foram possíveis pelos conhecimentos realizados com a minha participação em Conferências internacionais. Em particular, agradeço a Francesco Serti, Fergall McCann, Chiara Criscuolo, Daniela Maggioni e Gabor Békes. Agradeço finalmente à Fundação para a Ciência e Tecnologia, ao Gabinete de Planeamento, Estratégia, Avaliação e Relações Internacionais do Ministério da Ciência Tecnologia e Ensino Superior, ao Instituto Nacional de Estatística e à Faculdade de Economia do Porto respectivamente, pelo apoio financeiro, estatístico e técnico prestado. iii Sumário A presente tese centra-se no estudo do papel que o envolvimento das empresas portuguesas no comércio internacional pode ter para a respectiva capacidade inovadora e de melhoria de produtividade. O principal objectivo desta tese é o de contribuir para o esforço de clarificação, notório na literatura científica microeconómica nos últimos dez anos, acerca do nexo de causalidade entre o comércio internacional, por um lado, e a produtividade e inovação, por outro lado. Estudando empresas portuguesas entre 1996 e 2003, os resultados obtidos são importantes para o referido propósito de clarificação. Após um capítulo introdutório, a tese compreende mais cinco capítulos. No segundo capítulo desta tese apresentamos uma revisão de literatura sobre a hipótese de que as empresas aprendem através das suas exportações bem como sobre conceitos relacionados, como é o caso do tema da selecção das empresas para a actividade exportadora. Nesse sentido, apresentamos os principais estudos existentes, salientando as metodologias usadas e as principais conclusões de uma década de investigação sobre este assunto. No terceiro capítulo usamos dados do quarto Inquérito Comunitário à Inovação para Portugal e testamos a hipótese segundo a qual as empresas portuguesas mais inovadoras são as exportadoras e as filiais ou subsidiárias de empresas multinacionais estrangeiras. Através da estimação de várias Funções de Produção de Tecnologia, obtivemos resultados que sugerem a confirmação de tal hipótese. No quarto capítulo produzimos uma derivação simples do capítulo anterior para testar, em ligação com a internacionalização das empresas, o papel da Gestão dos Recursos Humanos sobre a aptidão inovadora das firmas portuguesas. Os resultados obtidos provam a importância dessa gestão para a capacidade inovadora das empresas. No quinto capítulo utilizamos dados económico-financeiros e sobre o comércio internacional da mais ampla amostra disponível de empresas portuguesas (1996-2003) visando avaliar a importância, a concentração e o papel do comércio internacional sobre o desempenho global das empresas e em especial sobre a sua produtividade. Uma forte correlação foi detectada entre essas variáveis. No sexto capítulo, usando a mesma base de dados e utilizando sobretudo técnicas de Matching e estimadores de diferenças às diferenças, confirmamos, em termos globais para as empresas portuguesas, quer a hipótese da selecção das melhoras empresas para os mercados exportadores, quer a hipótese da aprendizagem através das exportações. iv Summary The present thesis focuses on the role of international trade engagement for Portuguese firms’ ability to innovate and to increase productivity. Through this research we intend to contribute to the effort of clarification, which has span the literature over the last ten years, about the causality nexus between trade, on the one side, and productivity and innovation, on the other side. By taking into consideration the Portuguese case between 1996 and 2003, we consider that the outcomes of our work are important for the referred clarification. After an introductory chapter, the thesis comprises more five chapters. In the second chapter we review the literature on learning by exporting argument and connected issues as self-selection to exports. We present the existing studies, the methodologies used and the main conclusions of a decade of investigation on this issue, which relates firms’ efficiency and their respective international trade involvement. In the third chapter, we use data on the fourth Community Innovation Survey for Portugal and we test the hypothesis that most innovative Portuguese firms are exporters and affiliates/subsidiaries of foreign multinationals. By estimating several Knowledge Production Functions, we obtained results that suggest the validity of such hypothesis. In the fourth chapter, we produce a simple derivation of the previous chapter in order to test both the role of Human Resources Management and of internationalization for the innovativeness of Portuguese firms. We found the role of personnel training highly significant for the innovative process of Portuguese firms. In the fifth chapter, we use economic, financial and international trade data for a large sample of Portuguese firms with the aim of confirm the importance, concentration and role of international trade for the performance of firms. A strong correlation is observed across these variables. In the sixth chapter, using the same dataset as before and mainly Matching techniques and difference in differences estimators we confirm the learning by exporting hypothesis for Portuguese firms in the period 1996-2003. Moreover, the self selection hypothesis is also validated. v Contents Nota biográfica Agradecimentos ii iii Sumário Summary Contents List of tables iv List of abbreviations xv CHAPTER 1 INTRODUCTION 1 CHAPTER 2 LEARNING-BY-EXPORTING: WHAT WE KNOW AND WHAT WE WOULD LIKE TO KNOW 5 2.1. Introduction 6 2.2. Conceptual issues 7 2.2.1. Learning-by-exporting: origins of the concept 7 2.2.2. Connections with similar concepts 9 v vi x 2.2.2.1. Learning-by-doing 9 2.2.2.2. Increases in capacity utilisation 10 2.2.2.3. Productivity spillovers of exports 10 2.2.2.4. Learning-to-export 11 2.2.2.5. Learning to innovate (by exporting) 12 Fundamental causes of LBE 12 Review of empirical literature on LBE 13 2.3.1. Modelling empirical work 13 2.3.2. Methodological issues 14 2.2.3. 2.3. 2.3.2.1. Case studies 15 2.3.2.2. Micro-panel data studies 15 2.3.3. Review of empirical studies 20 2.3.4. The misestimation of LBE 26 2.3.4.1. Underestimation of LBE 26 2.3.4.2. Overestimation of LBE 28 2.4. 2.4.1. Trade and productivity revisited 28 The importance of imports 28 vi 2.4.2. Beyond within-firm level: reallocation effects across firms 30 2.4.3. Further investigation lines 31 2.5. Concluding remarks 32 CHAPTER 3 WHICH PORTUGUESE FIRMS ARE INNOVATIVE? THE IMPORTANCE MULTINATIONALS AND EXPORTERS MOST OF 33 3.1. 3.2. 3.3. 3.4. Introduction Innovation factors and empirical literature International factors of innovation: theory and modelling Empirical studies on innovation and foreign exposure Data issues on innovation in Portuguese firms: a summary Estimation issues and econometric strategies 34 36 36 38 39 46 3.5. Estimation results 47 Estimation of Knowledge Production Function for “Innovation – 48 3.2.1. 3.2.2. 3.5.1. Product or Process” 3.5.2. Estimation of Knowledge Production Function for “Innovation 50 Protection” 3.5.3. Estimation of Knowledge Production Function for “Novel Sales” 51 3.6. Innovation accounting 53 3.7. Concluding remarks 55 CHAPTER 4 THE IMPORTANCE OF HUMAN RESOURCES TO THE INNOVATIVE CAPACITY OF PORTUGUESE FIRMS 57 4.1. Introduction 58 4.2. Human capital approach to firms´ innovativeness 58 4.3. The particular importance of training for the innovative ability of 60 firms 4.4 The Knowledge Production Function methodology 61 4.5. Data issues, variables used and summary statistics 62 4.6. Hypotheses to test and results achieved 64 4.7. Concluding remarks 68 vii CHAPTER 5 ECONOMIC PERFORMANCE AND INTERNATIONAL 69 TRADE ENGAGEMENT: THE CASE OF PORTUGUESE MANUFACTURING FIRMS 5.1. Introduction 70 5.2. Data description 72 5.3. Evidence of firms’ heterogeneity in relation to international trade 73 5.3.1. International-trade propensity, intensity and persistency 73 5.3.2. International trade concentration 75 5.3.2.1. Internationalization for few firms 75 5.3.2.2. Concentration of international trade: intra- and inter-sectors 77 5.3.2.3. Concentration along the extensive margin 78 5.3.2.4. Concentration along the intensive margin 81 5.3.3. International trade status persistency 82 Measuring traders’ premium 83 5.4.1. Trader status 83 5.4.2. Trader extensive margins 87 5.4.3. Trader market heterogeneity 91 5.4. 5.4.3.1. Assessing traders’ heterogeneity 91 5.4.3.2. Robustness checks 96 5.4.4. 5.5. Dynamic specification 100 Conclusions 101 CHAPTER 6 SELF-SELECTION AND LEARNING-BY-EXPORTING IN 103 PORTUGUESE MANUFACTURING FIRMS 6.1. Introduction 104 6.2. Data description 106 6.3. Self-selection to exports 107 6.3.1. Self-selection reviewed 107 6.3.2. Self-selection tested 108 Post-entry effects 118 Post entry effects assessed by Fixed Effects model 119 6.4. 6.4.1. viii 6.4.2. Post entry effects assessed by Matching methods 121 6.4.3. Estimating the propensity score matching 125 6.4.4. Assessing matching quality 131 6.4.5. Propensity Score Matching with Differences in Differences 133 estimator 6.4.6. Learning channels and detailed Learning-by-Exporting analysis 140 Concluding remarks 153 BIBLIOGRAPHY 156 A APPENDIX TO CHAPTER 3 176 B APPENDIX TO CHAPTER 5 179 C APPENDIX TO CHAPTER 6 182 6.5. ix List of tables 3.1 Summary statistics on innovation performance 41 3.2 Knowledge outputs (mean values) 42 3.3 Knowledge inputs (mean values) 43 3.4 Knowledge flows (mean values) 44 3.5. Knowledge output productivity 45 3.6. 49 3.7 Estimates of Knowledge Production Function for “Innovation – Product or Process” Estimates of Knowledge Production Function for “Innovation Protection” 3.8 Estimates of Knowledge Production Function for “Novel Sales” 52 3.9 Innovation accounting statistics 55 4.1 66 4.2 Estimates of Knowledge Production Function for “Innovation – product or process”. Personnel Training detached Obstacles to innovation for non-innovative firms 5.1 International trade participation rate 74 5.2 Distribution of Portuguese exporters by export intensity levels 74 5.3 Export intensity of all exporting firms and of persistent exporting firms 75 5.4 Importance of “superstar firms” 76 5.5 Dimension of firms and of exporting firms 76 5.6 Concentration of Portuguese firms’ employees, sales and trade (1996 and 77 2003) Concentration of Portuguese firms (average 1996-2003) 78 5.7 51 67 5.9.1 Elasticity of the number of firms with respect to the number of country 79 destinations and goods exported Elasticity of the number of firms with respect to the number of country 79 origins and goods imported Distribution of export firms (2003) by number of goods and destinations 80 5.9.2. Distribution of export values (2003) by number of goods and destinations 5.8.1 5.8.2 x 80 5.10.1 Distribution of import firms (2003) by number of goods and destinations 81 5.10.2 Distribution of import values (2003) by number of goods and destinations 81 5.11 Trade status transition matrix from 1996 to 1999 82 5.12 Trade status transition matrix from 2000 to 2003 83 5.13 Trading status different average performances, 1996-2003 84 5.14 85 5.15 Firm heterogeneity and internationalized status, Pooled OLS (1996-2003) Firm heterogeneity and internationalized status, FE (1996-2003) 5.16 TWT coefficients for TFP and each group of firms 87 5.17 LP and TFP superiority (%) of TWT with high extensive margins 88 5.18 5.19 Firm heterogeneity along sector and country extensive margins, Pooled 89 OLS Firm heterogeneity along sector and country extensive margins, FE 90 5.20 TWT coefficients for TFP and each group of firms 90 5.21 Exporter’s different average performance, 1996-2003 92 5.22 Importers’ performance differences (1996-2003) 92 5.23 Trade premium by type of country development, 1996-2003; Pooled OLS Trade premium by type of country development, 1996-2003; FE Model Trade premium of TFP by type of country: sectoral analysis, 1996-2003 Trade premium by type of country development, 1996-2003 (Exports to Spain detached) Table 5.27.: Trade premium by type of country development, 1996-2003; Pooled OLS. (Exports to Spain and imports from Germany detached) Trade premium by type of country development: 1996-2003; FE model (Exports to Spain and imports from Germany detached) Fearless firms’ superiority 94 5.24 5.25 5.26 5.27 5.28 5.29 5.30 5.31 86 95 96 97 97 98 98 Trade premiums by type of country development, 1996-2003; Pooled OLS 99 (Exports to Spain and to Difficult Countries are detached) Trade premiums by type of country development, 1996-2003; FE model 99 (Exports to Spain and to Difficult Countries are detached) xi 5.32 Trade premium of TFP by type of country: sectoral analysis, 1996-2003 100 5.33 101 6.1 Dynamic panel data model for equation 5.1; Dependent variable: ln TFP Export Starters 6.2 Self selection to export (probit model) 109 6.3 Self selection to export using import status (probit model) 110 6.4 Self-Selection: levels 111 6.5 Self-Selection: growth-rates 112 6.6 6.7 Import share trend of starters and of never exporters before and after 114 exports begin Growth of imports (%) for export starters 114 6.8 Self selection by destination country of exports 116 6.9 Self-Selection in levels for different groups of sectors 117 6.10 Self-Selection in wage levels for all firms and Group 2 firms 118 6.11 Learning effects of entry to exports on TFP, using FE model 120 6.12 Number of starters upon each definition 125 6.13 Number of controls upon each definition 125 6.14a 132 6.15 Assessing the matching quality Comparison between treated and control at t-1 Assessing the matching quality Comparison between treated and control at t-2 ATT effects: PSM-DID estimations (raw version) 6.16 ATT effects: PSM-DID estimations (raw version) 136 6.17 137 6.20 ATT effects: PSM-DID “fine” estimations (PSM performed cross-section by cross section and year by year) PSM-DID (TFP) firms splitted by consecutive number of years observed after “start” PSM-DID (TFP) firms splitted by the number of years observed before “start” ATT effects: PSM-DID estimations of yearly growth rates 6.21 Export and import starters 141 6.14b 6.18 6.19 xii 109 132 135 138 139 140 6.22 ATT effects: PSM-DID estimations on TFP controlling for import starters 141 6.23 ATT effects: PSM-DID estimations with different import status 142 6.24 6.25.a ATT effects: PSM-DID estimations of TFP according to starters 144 comparative advantage or disadvantage PSM-DID estimations of TFP according to sectoral groups 145 6.25.b PSM-DID estimations of TFP according to CAE sectors 6.26 ATT effects: PSM-DID estimations according to starters foreign capital or 147 skilled labour ATT effects: PSM-DID of TFP estimations according to starters size 148 6.27 146 6.29 ATT effects: PSM-DID estimations of TFP according to starters export 149 intensity in the first three years after exporting begins ATT effects: PSM-DID estimations of TFP according to TFP initial level 150 6.30 ATT effects: PSM-DID estimations of TFP according to initial wage level 150 6.31 ATT effects: PSM-DID (TFP) according to exports´ destinations 151 6.32 ATT effects: PSM-DID for TFP according to the entry year of starters 153 A.1 Survey Questions in CIS 4 176 A.2 177 A.3 Estimates of Knowledge Production Function for Sales Share of Novel Sales Innovation Accounting for GM firms and Innovation - Product or Process B.1 Sectoral Theil Index 179 B.2 Between sectors concentration of exports and Trade participation rates 180 B.3 Export intensive margin 180 B.4 Export growth (1996-2003) to the 10 most frequent destinations 181 B.5 Toughest markets for exports (Difficult Countries – DC) 181 C.1 Granger Causality tests between ln TFP and Export Ratio 182 C.2 C.3 Percentual differential between the weight of each industrial sector in 182 export starters and in all exporters (1997-2002) Number of repeated starters 182 C.4 Treated and controls using the more restrict concept: 6.28 xiii 178 183 C.5 Treated and controls using the intermediate concept: 184 C.6 ATT effects: PSM-DID estimations; covariates at t-2 184 C.7 Relative specialization of Portuguese exports (1995-2004); Balassa Index 185 for Portuguese industries (average for 1995-1999 and 2000 -2004) Comparative level of Starters in each year 185 C.8 xiv List of abbreviations ATT Average Treatment effect on the Treated BP Breusch-Pagan test CAE Código de Actividade Económica – sectoral classification of business activities CI Capital Intensity CIA Conditional Independence Assumption CIS Community Innovation Surveys CSA Common Support Assumption DC Difficult Countries DID Differences in Differences DOM Domestic firms Dsmall Small Firms E_EU Exports only to European Union E_EU+ODev Exports only to EU and ODev E_EU+PL Exports only to EU and PL E_Multiple Exports to all other possible combinations of markets E_NDev Exports only to Non-Developed countries E_Odev Exports only to other Developed countries E_PL Exports only to PL countries E_SPA Exports only to Spain ECE Estatísticas do Comércio Externo - external-trade information EU European Union EUROSTAT European Statistical Office of the European Union EXP Exporters FDI Foreign Direct Investment FE Fixed-Effects models FF Fearless Firms Forcap Foreign share of Capital GM Global Multinationals GMM General Method of Moments xv GPS Generalised Propensity Score Gr 1 Group 1 Gr 2 Group 2 Gr 3 Group 3 Gr 4 Group 4 Gr 5 Group 5 HHS Health, Hygiene and Security in work HMG Home-market Growth HR Human Resources HRM Human Resources Management I_Dev Imports only from ODev countries I_EU Imports only from EU countries I_Ger Imports only from Germany I_Multiple Imports from other possible combinations of countries I_NDev Imports from NDev countries I_PL Imports only from PL countries IAE Inquérito Anual às Empresas – annual balance-sheet information IM Internal Multinationals Imp firm Imports ImpMac firm Imports Machines IND Industry Resources INE Portuguese National Statistics Institute IPPOM Innovation on Product, Process, Organization and Marketing ISGEP International Study Group on Exports and Productivity K knowledge stock KPF Knowledge Production Function LBD Learning-by-Doing LBE Learning-by-Exporting LBI Learning-by-Importing LCR Labour Costs Reductions by unit of goods produced or services provided LIBE Learning-to-Innovate-by-Exporting ln logarithm LP Labour Productivity NACE Statistical Classification of Economic Activities in the European Community NCE Number of Countries where Exports are sold xvi NCI Number of countries from where Imports are bought NDev Non Developed countries NSE Number of products / Sectors Exported NSI Number of products / Sectors Imported NT Non-Traders OCES Observatório da Ciência e do Ensino Superior ODev Other Developed countries OE firms Only Exporting OECD Organization for Economic Co-Operation and Development OI firms Only Importing OLS Ordinary Least Squares PL Portuguese speaking countries Post_CA post-entry period for starters in Comparative Advantage (CA) sectors Post_CD post-entry period for starters in Comparative Disadvantage sectors (CD) PSE Productivity Spillovers by Exporting PSM-DID Propensity Score Matching with Differences in Differences estimators R&D Research and Development RI Internal Resources Skill firm that has Skilled production employees SS Self-Selection TFP Total Factor Productivity TWT Two-Way Traders U.S. United States of America UK United Kingdom ULC Labour Cost per unit of sales (ULC) XMG Export-Market Growth xvii